Spatial Hearing in Audio Practice

Spatial hearing informs how we capture, shape, and evaluate sound scenes:

• Microphone technique and stereo array selection (localization cues, stereo width)
• Panning, depth, and spatial processing (level/time differences, reverberation cues)
• Immersive and binaural production workflows
• Assessment of monitoring and room acoustics (imaging, translation)

Spatial references in binaural hearing

• Azimuth (horizontal angle)
• Elevation (vertical angle)
• Distance

Anatomical reference planes: frontal (front/back), horizontal (upper/lower), median (left/right)

Lateralization and localization

Binaural hearing and source localization

The first theory of lateral sound localization was proposed by Lord Rayleigh as "duplex theory." Later extended with Blauert's spectral cues.

• Interaural Time Difference (ITD):
  • Difference in arrival time between ears. Effective for low frequencies (<1500 Hz) because long wavelengths maintain phase coherence between ears.
• Interaural Level Difference (ILD):
  • Difference in sound intensity between ears. Effective for high frequencies (>1500 Hz) due to head shadow effect (short wavelengths are blocked by the head).
• Spectral Cues:
  • Additional information from how the head and pinnae filter sound (HRTF).

Interaural time difference (ITD)

Interaural time difference (ITD)

ITD helps localize low-frequency sounds.

For an ear distance of cm, the natural interaural time delay is approximately

• Phase delays at low frequencies (if wavelength is greater than half the distance between the ears )

▶ ITD noise bursts (listen via headphones)

Interaural level difference (ILD)

Interaural Level Difference (ILD), also known as Interaural Intensity Difference (IID), plays a key role in the localization of high-frequency sounds.

• Low frequencies bend around the head with minimal attenuation
• High frequencies are significantly attenuated due to the head shadow effect
• Level differences increase above ~1600 Hz

▶ ILD noise bursts (listen via headphones)

Interaural level difference (ILD)

ITD and ILD

Localization accuracy:

• 1° for sources in front
• 15° for sources to the sides

Frequency ranges:

• ITD dominant for Hz
• Both contribute between 1000-1500 Hz
• ILD dominant for Hz
• No localization below 80 Hz

White noise at -90° azimuth: maximum ITD of 0.63 ms

Cone of confusion

• Front-back ambiguity: Cannot discriminate between sounds originating from the front or rear.
• No elevation cues: ITD and ILD do not provide information about vertical localization (elevation).

Resolving ambiguity

• Tilting the head: Introduces new timing and level information to help resolve the location of the sound source.
• Spectral cues: The filtering effects of the pinnae and torso shape the sound spectrum, helping to distinguish elevation and front-back positioning.

Spectral cues

The outer ear, auricle (pinna), and ear canal act as a resonator system, shaping sound based on its direction of incidence.

• Frequency-dependent filtering provides spatial information, aiding in sound localization.
• Encodes vertical and front-back localization cues (median plane), primarily through frequency bands affected by the pinna.

Directional bands (after J. Blauert)

Head related transfer function (HRTF)

HRTF describes how spatial audio cues are encoded in the sound reaching the ears, allowing for sound source localization.

The torso, head, and pinna act as direction-dependent filters, introducing frequency-specific alterations to the sound.

→ This effect can be mathematically represented as a transfer function.

▶ Clicks rotating 360° up - down
(listen via headphones)

HRTF measurement

Applications of HRTF

• 3D audio rendering
• VR and AR
• Binaural recording

→ The auditory system adapts to a modified head-related transfer function over time.

Franssen effect

A live demonstration of how the onset of a sound can dominate its perceived location.

• A pure tone begins abruptly at one speaker.
• The signal then moves gradually to the opposite speaker.
• The sound may still appear to remain at its initial location.

Requires separated stereo speakers in a reverberant room.
The effect does not work with headphones.

See previous slide before playing.

▶ Play Franssen effect

Auditory distance perception

The auditory system has limited ability to determine distance, relying on:

• Initial Time Delay Gap (ITDG): Time between direct sound and first reflection
• Direct-to-reverberant ratio: Closer sources have more direct sound
• Reverberation density: More diffuse reflections indicate greater distance
• High-frequency absorption: Distant sounds lose high-frequency content
• Loudness: Closer sources perceived as louder
• Motion parallax: Closer sources shift position faster for moving listeners

▶ Distance close - far

Stereophonic Sound

When two channels are played through separate speakers, listeners perceive a soundstage extending between those speakers.

• Summing localization: Brain combines signals from both speakers to create phantom sound sources
• Phantom center: Virtual sound source perceived between the speakers

Creates an illusion of multi-directional spatial perspective.

Artificial stereo and phantom center

Dual-mono signal

Mono source material played back through two stereo channels with identical signals on both the left and right channels.

• Identical waveform in both L and R channels
• Creates phantom center image when played through speakers
• Perceived as center image in headphones (lateralization)
• Common in broadcast, mono recordings, and centered mix elements

Summing localization

Two speakers create a phantom sound source between them by manipulating the same binaural cues the brain uses for natural localization:

• Amplitude panning (ILD)
• Time delay (ITD)

Summing localisation · data redrawn after Karl Wendt (1963)

Stereo recording

ITD and ILD are used in common stereo recording practices:
Stereo recording techniques:

• Coincident pair (X-Y): Accurate localization
• Spaced pair (A-B): Wide stereo image

Summary: sound localization

• ITD (time differences) for low frequencies
• ILD (level differences) for high frequencies

Ambiguity (Cone of confusion) resolved through head movement and spectral cues (HRTF) for elevation and front-back.

Applications:

• Stereo reproduction relies on phantom imaging
• HRTF measurements enable binaural 3D audio